PixLab provides Think-Act and Vision Workspace alongside OCR, vision-language, document-processing, and DocScan APIs—helping teams turn visual and document content into useful, reviewable results.
Start with the desktop agent, use Vision Workspace in the browser, or integrate the underlying APIs directly.
Think-Act product screenshot
Think-Act coordinates desktop work. The surrounding PixLab services handle document parsing, supported identity documents, browser-based analysis and programmatic vision tasks.
Bring in context
Screenshots & images
Visual context already on the desktop
Documents & scans
PDFs, receipts, forms, contracts
Recorded processes
Walkthroughs and bug reproductions
Selected or supplied text
Drafts, notes, messages, and instructions
Agentic Center
Understand the relevant context, plan the next step, use enabled capabilities, pause when approval is required and keep supported activity visible for review.
Context
Source material
Scope
Enabled tools
Gate
Approval policy
Review
Visible activity
Products can be used independently. The value of the platform is that the same document and vision capabilities can support desktop workflows, browser workflows and developer-built applications.
Try Think-Act NowRewrite text directly, inspect a document, or record a process on its own. Move into the Desktop Agent when a task needs several capabilities or connected tools to work together.
Desktop Agent
Give Think-Act a clear outcome, relevant files or screenshots and any limits it should follow. It uses the capabilities available to the run and keeps supported activity visible.
A run can include
Writing Assistant
Start with selected or copied text, a document or another checked source, then review the resulting draft before using it.
Explore writing assistance →OCR & document intelligence
Work with supported screenshots, scans, image-based PDFs, receipts, invoices, contracts and forms, then review extracted text or fields against the source.
Explore document intelligence →Screen Recorder
Record a repeated desktop process with optional narration, then use the reviewed recording as context for an SOP, checklist or bug report.
Explore screen recording →Reusable local skills
Save a reviewed workflow with its prerequisites and expected result when repetition is worthwhile. A skill can guide future runs, but it cannot bypass missing tools or permissions.
Explore skills & integrations →Think-Act's agent works through bounded steps. You provide the task and material, the agent uses only the tools enabled for that run, pauses when approval is required, and delivers visible results you can inspect before deciding what happens next.
Agent workflow screenshot
Input
Give Think-Act a specific outcome, the documents it should use, and a clear stopping point. For example: compare an invoice with its purchase order and list anything that needs review.
Understanding
OCR and document tools can make scanned text and fields usable for the task. Think-Act can then compare the relevant values and keep the source material in context.
Example: compare totals, tax, dates, vendor details, and line items.
Execution
The Desktop Agent uses only the capabilities available to that run. If a required tool or connection is unavailable, the workflow should surface that limitation instead of silently assuming access.
Human decision
When the configured policy requires approval, Think-Act pauses before supported actions such as changing a file, sending information, or calling an external system. Each action can be allowed, confirmation-required, or blocked.
Review
Inspect the comparison, supporting observations, and workflow log. Confirm important differences against the invoice and purchase order before using the result or starting another action.
PixLab's LLM Parse Service transforms raw documents into structured, machine-readable content ready for retrieval-augmented generation, custom AI applications, and agent tasks. It handles multi-format inputs and produces output directly compatible with LLM frameworks.
Supported formats: PDF, DOCX, PPTX, XLSX, HTML, JPEG, EPUB, and more
PixLab DocScan is an SDK-free REST API for scanning and extracting structured identity data from passports, driver's licences, national IDs, visas, and residence permits across 200+ countries and territories. It returns JSON you can use directly in KYC, onboarding, and document verification workflows.
200+ countries & territories
Including MRZ and non-MRZ documents
Privacy first
In-memory processing, no image retention
Structured JSON output
Name, DOB, document number, expiry, and more
Any language, no SDK
Plain REST — works with any HTTP client
{
"type": "PASSPORT",
"face_url": "https://s3.amazonaws.com/media.pixlab.xyz/24p5ba822a00df7f.png",
"mrz_img_url": "https://s3.amazonaws.com/media.pixlab.xyz/24p5ba822a1e426d.png",
"mrz_raw_text": "P<UTOERIKSSON<<ANNA<MARIA<<<<<<<<<<<<<<<<<<<<<<<\nL898902C36UTO7408122F1204159ZE184226B<<<<<10",
"fields": {
"issuingCountry": "UTO",
"fullName": "ERIKSSON ANNA MARIA",
"documentNumber": "L898902C3",
"checkDigit": "6/2/9/1",
"nationality": "UTO",
"dateOfBirth": "1974-08-12",
"sex": "F",
"dateOfExpiry": "2012-04-15",
"personalNumber": "ZE184226B",
"finalCheckDigit": "0"
},
"status": 200
}
Vision Workspace is PixLab's browser-based productivity suite. Extract data from PDFs, images, and spreadsheets. Chat with documents using RAG-powered conversation. No installation required.
Useful agentic AI does not need to begin with a giant autonomous process. Start with bounded document, writing and process workflows where the source and expected result are clear.
Extract searchable text, summarize relevant sections, and keep the source available for verification.
Document workflows →Pull required fields into a table or machine-readable result and flag anything unclear instead of silently guessing.
LLM parsing tools →Prepare a clearer draft from selected text, a document, meeting notes, or other supplied context.
Writing assistance →Capture a workflow or reproduction, generate a draft procedure, and verify the documented steps against the recording.
Operations workflows →Document prerequisites, checks, and expected output, then save a proven method as a reusable local skill.
Skills and integrations →Use OCR, parsing, DocScan or vision-language APIs as explicit application steps.
Vision API reference →Think-Act limits automation through explicit tool availability, configurable action policies, and reviewable activity. The agent cannot use a tool you have not installed and enabled.
Enable only what the task needs
Connectors not available to a run cannot be used by it.
Allow, ask, or block per action
Set each supported action type according to its impact.
Review what the agent did
Inspect tool calls, results, and observations after a run.
Unavailable connectors cannot be used
They stay unavailable until the connection is restored and enabled.
Approval Controls Screenshot
Recommended starting policy
Read selected content
Create or change files
Send or publish
Delete or risky commands
Use PixLab's APIs for document parsing, OCR, layout understanding, structured extraction, DocScan, and vision-language workflows when your own product needs to process visual or document input programmatically.
Segment, chunk, and embed documents. Feed clean, structured content into your LLM framework or retrieval pipeline.
Learn more about LLM parsing →Query images and documents with Vision Language Models. Extract structured data, summarize content, and build visual understanding into your product.
Learn more about Vision APIs →Scan and extract structured JSON from supported ID documents using a single HTTPS call. No SDK. Works from Python, Node.js, PHP, Java, Go, and any other HTTPS-capable language or framework.
Learn more about DocScan →Clear answers about product roles, human control, integrations, and developer access. Can't find the answer you need? Contact support.
Think-Act is PixLab's Windows desktop AI workspace for writing assistance, OCR and document review, screen recording, local skills, and bounded multi-step tasks. It is designed to keep work visible and reviewable—not to operate as an uncontrolled autonomous agent.
Think-Act is the desktop agent at the center of the platform. PixLab's LLM and document-parsing APIs prepare and structure content, DocScan processes supported ID documents, Vision Workspace provides browser-based tools, and Vision APIs bring image and document understanding into applications. See how these parts fit together in the connected platform overview.
You define the task and choose which supported tools are available. Configured action policies can allow an action, require confirmation, or block it. Think-Act also provides workflow activity and results for review, and important output should be checked before use. Read more about Think-Act security and approval controls.
Not for every task. Writing assistance, OCR and document review, screen recording, and local skills can be useful without external connectors. A workflow that needs another application or service requires the relevant supported integration to be installed, connected, and enabled. Review the difference between local skills and integrations.
Use Vision Workspace when you want browser-based OCR, document processing, document chat, and vision tools without a desktop-agent workflow. Use Think-Act when the work belongs in a bounded, reviewable desktop task that may span enabled tools.
DocScan is PixLab's REST API for scanning supported ID documents and returning normalized fields as structured data. It is intended for identity-document capture and extraction workflows, not general-purpose document parsing.
PixLab provides APIs for LLM-ready document parsing, OCR, document segmentation, structured extraction, DocScan, and vision-language analysis. These services can be used independently of the Think-Act desktop product. Start with the Vision API reference or explore the LLM and parsing APIs.
Choose your next step
Use Think-Act for controlled desktop work, Vision Workspace for interactive document workflows, or PixLab's APIs when the capability belongs inside your application.